Rapid taxonomic categorization of short, abundant virus sequences for
ecological analyses
Abstract
Public health concerns about recent viral epidemics have motivated
researchers to seek novel ways to understand pathogen infection in
native, wildlife hosts. With its deep history of tools and perspectives
for understanding the abundance and distribution of organisms, ecology
can shed new light on viral infection dynamics. However, datasets
allowing deep explorations of viral communities from an ecological
perspective are lacking. We sampled 1,086 bats from two, adjacent Puerto
Rican caves and tested them for infection by herpesviruses, resulting in
3,131 short, viral sequences. Using percent identity of nucleotides and
a machine learning algorithm (affinity propagation), we categorized
herpesviruses into 43 operational taxonomic units (OTUs), to be used in
place of species in subsequent ecological analyses. Herpesvirus
metacommunities demonstrated long-tailed rank frequency distributions at
all analyzed levels of host organization (i.e., individual, population,
and community). Although 13 herpesvirus OTUs were detected in more than
one host species, OTUs generally exhibited host specificity by infecting
a single core host species at a significantly higher prevalence than in
all satellite species combined. We describes the natural history of
herpesvirus metacommunities in Puerto Rican bats and suggest that
viruses follow the general law that communities comprise few common and
many rare species. To guide future efforts in the field of viral
ecology, hypotheses are presented regarding mechanisms that contribute
to these patterns.